OpenAI Launches Shopping Research in ChatGPT

OpenAI has started rolling out shopping research, a new ChatGPT experience that researches products across the internet and produces a personalized buyer's guide in minutes. The feature is available starting today on mobile and web for logged-in users on Free, Go, Plus, and Pro plans, with nearly unlimited usage through the holidays.

The stakes are significant: OpenAI says hundreds of millions of people already use ChatGPT to find, understand, and compare products, and the new tool moves the company deeper into commercial search territory that Google and Amazon currently dominate.

How it works for users

Users describe what they want in plain language — OpenAI's examples include "Find the quietest cordless stick vacuum for a small apartment" and "I need a gift for my four year old niece who loves art." The tool asks clarifying questions about budget, recipient, and features, then researches the web for up-to-date prices, availability, reviews, specs, and images. Users can steer the process by marking suggested products as "Not interested" or "More like this."

After a few minutes, the user receives a buyer's guide with top products, key differences, tradeoffs, and current information from retailers. OpenAI frames it as replacing manual comparison work: "It's a clear summary that normally would take a lot of comparing, reading, and checking on your own."

The feature performs especially well, according to OpenAI, in detail-heavy categories: electronics, beauty, home and garden, kitchen and appliances, and sports and outdoor. For simple queries like a price check, a regular ChatGPT response remains faster.

The model behind it

Shopping research runs on a version of GPT-5 mini trained with reinforcement learning specifically for shopping tasks. More precisely, it is a mini model post-trained on GPT-5-Thinking-mini. OpenAI created a new evaluation built from difficult product discovery queries with many constraints, measuring product accuracy as the percentage of products in each response that meet user requirements such as price, color, material, and specs.

OpenAI trained the model to read trusted sites, cite reliable sources, and synthesize information across many sources. The interactive design lets it update research in real time based on new constraints and user feedback.

Memory, Pulse, and commerce

Shopping research builds on ChatGPT memory and past conversations for personalization. If ChatGPT knows a user is into gaming, it can factor that into a laptop recommendation.

The feature is also available in ChatGPT Pulse, currently limited to Pro users. Pulse can proactively suggest buyer's guides based on past conversations — someone who discussed e-bikes might see a card suggesting accessories.

Purchases currently require clicking through to the retailer's site. In the future, users will be able to buy directly through ChatGPT from merchants enrolled in Instant Checkout, where available — a step that would turn ChatGPT from a research tool into a transaction channel.

Transparency and limits

OpenAI says chats are never shared with retailers, and results are organic, drawn from publicly available retail sites while avoiding low-quality or spammy sources. Merchants can request inclusion through an allowlisting process.

The company acknowledges the model is not perfect. It may make mistakes about product details like price and availability, and OpenAI encourages users to visit merchant sites for accurate details.

OpenAI says shopping research is just the start, with plans to expand category coverage, improve preference understanding, and add more intuitive ways to compare and discover products.